Assessing personality with a structured interview: The effect of faking and question type on interviewer ratings.

摘要

Research indicates that certain personality dimensions are related to both productive and counterproductive work behaviors. Studies also show that applicants commonly distort their responses on traditional paper and pencil personality inventories, and that this distortion can degrade test validity and reliability and ultimately influence personnel decisions. This study evaluated a structured interview as an alternate method for assessing personality. Behavioral and situational interviews were developed to measure subdimensions of agreeableness, conscientiousness, and emotional stability. Experienced interviewers ( N = 52) administered the interviews to upper-level undergraduates and graduate students (N = 143) who were instructed to respond honestly or to make them self look like the best candidate for a customer service manager position. Interviewees also completed scales of the same three Big Five dimensions from the revised NEO Personality Inventory under honest and fake instructions. In addition, interviewers and two close acquaintances of each interviewee evaluated interviewee personality with the NEO. The analysis of the data revealed that the faking instructions influenced self-report NEO scores about four times more than they influenced the structured interview ratings. There was mixed evidence for the construct validity of the personality interviews. A series of multitrait-multimethod analyses yielded convergent validities that were significantly larger than the discriminant validities, and an exploratory factor analysis produced three factors that corresponded to the three personality dimensions the interviews were designed to measure. A confirmatory factor analysis, however, did not reveal a good fit for the 3-factor model of the interview ratings. Further, interviewer ratings of the structured interview questions were highly related to their NEO ratings of interviewees, but correlated much less with NEO ratings provided by interviewees and their close acquaintances. Interview type, behavioral versus situational, did not moderate faking or construct validity. The implications of these results for personnel selection are discussed.